Gigaspaces.com - Application Server
Home
  • Send to a friend
  • Print

Webinar Q&A - On Demand Middleware for the Enterprise - Coming Up in 2010

  • How is this new functionality related to XAP product?
  • The new functionality is part of the existing GigaSpaces XAP product
    editions and will be available to all XAP users in our upcoming releases
  • What type of applications can benefit from these functionalities?
  • This is relevant for all types of applications. This functionality
    provides ease of use, reduced administration overhead and cost
    efficiency. It enables applications to benefit dynamic environments as
    well as reduce
  • Does this functionality work without virtualization?
  • Yes. As noted in the previous answer, it’s relevant to all GigaSpaces
    applications. The benefit for cloud and virtualization users is the
    ability to register custom event handlers to dynamically expand or
    reduce the capacity of the cluster by allocating new resources on the
    fly.
  • What communication protocols are used between agents and agents+grid manager?
  • The communication protocol used is called LRMI (lightweight RMI) that
    is the basis for all network communication within GigaSpaces XAP. It is
    an optimized version of the standard Java RMI protocol on top of TCP/IP
    with support for automatic failover, failure detection and highly
    efficient serialization.
  • How can I reduce the latency overhead?
  • The main factor for latency is the network roundtrip associated with
    every call to the data grid. So when accessing the data grid multiple
    times within the same business transaction or when performing complex
    calculations that involve multiple accesses to the data grid, this can
    add up to a significant amount of time. To reduce this overhead, you
    can collocate the code that accesses the data with the data. That means
    that each data grid instances can also be used for code execution and
    not just serve requests for data. The code executes on the node, and
    accesses each of the data grid instance locally in memory speeds. The
    end result is then returned back to the client. If the business
    transaction needs to access data located on more than one node, it can
    direct the code to execute on each of the relevant nodes and then
    aggregate the results at the client side. The code can be static
    (compiled and deployed on the server) or dynamic, which means it is
    sent from the client to the data grid nodes when executed.